Jump to content

ChinRey

Advisor
  • Posts

    8,379
  • Joined

  • Last visited

Posts posted by ChinRey

  1. 2 hours ago, Thecla said:

    Ok I assumed you could have multiple UV maps.

    To clarify this, you can only have one named UV map. But you can effectively have multiple UV maps simply by reusing the same UV space for different faces.

    Take this pillar for example (it's only a demonstration so I didn't texture it, I just gave the different faces different colours):

    image.thumb.png.30eb217f57b9ae0ae6c0b57e6c846f57.png

    The UV mapping for the main coulm and the what-cha-call-it close to the top (red and brown in the picture) looks like this:

    image.png.1624b3894cabdc39f1368e2a3ecb561e.png

    The base (yellow and green):

    image.png.eef3994229ef8ceddc224e122e10a933.png

    and the capital (everything else):

    image.png.1ac62f0740e2d7f24893f1de85ef5833.png

    As you see, there's a lot of overlapping here:

    image.png.d3959c6891f35af02512815399b972f7.png

    But since they are different faces, they are textured separately and won't interfere with each other. So although it's nominally only one UV map, for all practical purposes there are three different ones.

    • Like 1
    • Thanks 1
  2. 47 minutes ago, OptimoMaximo said:

    You say correctly, "should". I'd remind you that what you have in the build windows is a texture asset, any changes made on one instance would be propagated to all instances, because the only asset type that actually uses instancing in SL is the texture asset.

    No, it's perfectly possible to have a texture/map modifier that only is applied to a single instance of the asset: store it as a prim property and apply it client side. SL already has several texture/map modifiers that are handled this way.

    • Like 1
  3. 6 hours ago, IvyTechEngineer said:

    The image on the left would be harder to texture with an image?

    I don't think it would make much difference if any at all but try both and see. Btw, you do have access to the beta grid so you can do test uploads without paying for them right?

    5 hours ago, Frionil Fang said:

    Subdividing the split part and unwrapping gets you the exact same texture area for the split part, so there's exactly the same texture detail available in those 16 quads as the original undivided one

    That's almost correct but there are no quads in SL, only tris and that can sometimes make a difference. Look at these two cones:

    image.png.3a865522d7d94d72a7fe2f7216d8cd7f.png

    The one to the left has tris going all the way from top to bottom, the one to the right are split in four segments vertically and you can see how that improves the UV mapping. Here they are in wireframe mode:

    image.png.0575c43007cf1a5455bf86847db0a840.png

    As I said, I do not think it will make a difference worth mentioning in this particular case but it's worth a try.

  4. 1 hour ago, Fluffy Sharkfin said:

    It should be a matter of simply providing a couple of extra checkboxes in the texture tab of the build window which when checked will invert the red or green channel of the normal map, this would allow users to flip a normal map horizontally and/or vertically and then invert the corresponding channel(s) to compensate for the reversed texture orientation.

    That's what the normal map's swizzle means: http://wiki.polycount.com/wiki/Normal_Map_Technical_Details

     

    1 hour ago, Rick Nightingale said:

    Wonder if it's worth putting in a feature request... might do later.

    Now is the only time there's even a hint of chance LL will process such a feature request. The chance increases slightly if you focus on the fact that it will make SL compatible with both Unity and UE.

    • Like 1
    • Thanks 1
  5. 10 hours ago, IvyTechEngineer said:

    Thanks for the feedback. So are you telling me that texturing in SL would be better for large objects?

    Yes and no.

    I've done a lot of ground meshes and I've found that 32x32 m is the largest surface a single non-repeating 1024x1024 texture can cover with acceptable resolution and even that is pushing it a bit. For a larger surface you either split it up into several faces (and load the viewer down with a lot of 1024s) or you use a repeating texture. I do both: custom baked textures for areas where the difference really matters and a standard tiled one for the rest of the ground.

  6. 15 hours ago, OptimoMaximo said:

    IMHO, what you're looking for is a bit too much

    I'm not sure but isn't this just a matter of changing the swizzle? If so, it should be easy to implement and it could be useful for some other purposes as well. The viewer already has code for inverting a channel of a texture even. Maybe that code could be reused for normal maps?

    The fact that no other game/VR engine has it isn't really an argument. Such a function is only really useful as part of an advanced in-world texture manipulating system and that's something only SL and opensim offer.

    • Like 2
    • Thanks 1
  7. 5 hours ago, Ceka Cianci said:

    Now I have to log out and log back in! \o/

    I missed it this morning when I logged in.. hehehe

    The splash page is cycling through several different variants and it's been a while since I saw the one I posted a screenshot of so you might be too late for that one. The only thing that stays the same is the googly eyes.

    • Thanks 1
  8. 14 hours ago, Love Zhaoying said:

    I don't get the title. Where are they aiming Second Life, is it a pointy object? Or are they aiming "for" Second Life, meaning aiming to take its users / business segment? Good luck with that!

    IMVU had a rather agressive campaign targetting SL users at the time this thread was started. That was two years ago though. It didn't last long and it didn't work.

    I'm not always against necroposting but it's not appropriate in this case so let's just put this thread to rest.

    • Like 1
  9. I thought I'd ask ChatGPT about it and here's its answer:

    Quote

    As an AI language model, I can provide insights and information about the current state of artificial intelligence and its capabilities. However, it's important to note that the scenario you described is not currently possible with existing AI technology.

    While AI has made significant advancements in fields like computer vision, natural language processing, and robotics, it is not capable of turning virtual avatars into real people. Avatars are digital representations of people created for use in virtual environments, and they do not possess physical bodies or the biological components that make up a real human being.

    Furthermore, the idea of "turning an avatar into a real person" raises ethical concerns around identity and consent. It's important to approach these discussions with caution and consider the potential implications of such technologies.

    In summary, AI is not currently capable of turning avatars into real people, and any such scenario raises important ethical considerations.

    When I asked if I can quote it on this, it added:

    Quote

    Yes, you may quote my response. Please ensure that it is clear that my response is based on the current state of artificial intelligence and its capabilities as of my knowledge cutoff date of September 2021. It's also important to note that AI research and technology are continually evolving, and future advancements may change our understanding of what is possible.

     

    • Like 1
  10. 3 hours ago, Thecla said:

    Just the nature of the beast. If we had a more usable "mouselook" experience in SL, things might be different.

    Oh yes! If I was allowed to change only one thing in Second Life, it would be to add a first person view option without all the other UI changes we get with mouselook. No other improvement would be even nearly as important to me as that.

    All is not lost though. It's fairly easy to adjust your camera position closer away from the birdseye view and clsoer to your avatar. Penny Patton made a really good tutorial how to do it: https://pennycow.blogspot.com/2012/01/improving-sl-camera-short-version.html

    My own settings are a bit different from Penny's and closer to Jo Yardley's: https://joyardley.wordpress.com/2012/04/14/camera-placement-in-second-life/

    • Thanks 1
  11. 41 minutes ago, xDancingStarx said:

    I just facepalm asking myself why you would ever revive a thread with this title, especially if they're of a contrary opinion (why did you not just create a thread like "why Second Life is not dying" which is what your post was about)

    Normally I would agree with you but for this topic it's good to have the historical perspective.

  12. 1 hour ago, Sid Nagy said:

    I think the Lab (and so SL too) came close to an end during what turned out to be the Sansar fiasco.
    But now there should be years left in SL's lifespan if no unexpected things happen.

    I think Sansar helped save SL. The big problem Second Life had during the 2008-2013 time period was the constant pressure from LL's owners and top management to rush through new "shinies" to stack on top of the rather shaky fundament. Sansar took off that pressure and the Lindens who remained with SL shifted their focus to repair and maintenance which is what they should have done much earlier.

    • Like 2
  13. On 3/9/2023 at 2:00 AM, LittleMe Jewell said:

    Wow, this might be the oldest resurrection ever done here

    Yes but that actually makes it interesting, especially with the 20th anniversary coming up. Ten years ago, that's halfway from the beginning to now. Think of all the other virtual worlds that have popped up and died since then. Sansar, High FIdelity, Meta (not quite dead but definitely a zombie)...

    You know what, I suspect the title was correct at that time. SL was in a bad shape in 2013 and LL had even started working on a replacement. But then Oz and Ebbe came along. Ebbe cleaned up LL's organisation. Oz was tasked with patching up SL to keep it going until Sansar was ready. Only he did such a good job that when Sansar inevitably folded, SL was stronger than it had been for many years. Yes, we can talk about all the things they didn't do or couldn't do and all the things they got wrong, it would be a long list. But there was also so much they got right and it was enough.

    SL is still dying btw but aren't we all?

    • Like 1
  14. 3 hours ago, Arduenn Schwartzman said:

    It's great, though, that edit rights have been pinpointed as a probable cause.

    No, you misunderstand me. "Somebody who has editing rights to the object" includes the owner. I've corrected my post to eliminate the misunderstanding.

    It's actually a good idea to lock permanently rezzed objects anyway, espexially if more than one person has editing rights to them. If somebody needs to edit the object, they can always unlock it.

  15. On 2/28/2023 at 12:32 PM, Arduenn Schwartzman said:

    Any thoughts/ideas on this phenomenon, outside of what already has been stated in the other thread and on the relevant JIRAs?

    Nothing new but I can repeat something I said in the old thread about it: the workaround is to lock the object.

    The alpha switch bug is caused by miscommunication between the server and the viewer of the owner (or somebody else with editing rights to the object). If an object is locked, nobody will have editing rights to it so the problem goes away.

    • Like 1
  16. 6 hours ago, arisaie said:

    I went with the baking route because I am not at all proficient in graphics software like photoshop(and I think I googled it and its not possible to randomly scatter objects on the texture with photoshop). Even if I stretch the UV island to bounds inblender and bake it like that it turns out pixalated.

    You should try to use a mid level graphics program like paint.net. It's far more user friendly than Photoshop and Gimp and it still has nearly all the functions you actually need for SL textures. I haven't used Photoshop for years myself since I'm allergic to pay-per-month software. I do use Gimp for operations paint.net can't do but I hardly ever need it - it doesn't even happen every year.

    You are right that a image editor won't scatter objects randomly for you but it's not really that difficult to do it manually, just make a bunch of copies of a layer with a single leaf or a small group of them and move, rotate and resize each layer. An additional advantage to doing it manually is that you can add some discreet perlin noise shading to each layer - an amazingly effective way to add life and depth to the texture. I'm forever grateful to JubJub Forder for teaching me that last trick.

    Any reasonably advanced image editor have to crucial functions that Blender lacks as far as I know. One is the option to color the transparent parts of the texture. The black outline of your leaves are caused by the color of the transparent pixels bleeding onto the opaque ones. If you look at the last picture in my post, where alpha mode is set to none, you see that the alpha color is green, similar to the leaves themselves.

    The other is that they offer a variety of scaling algorithms. There are several ways pixels can be combined or split when an image is resized and which method you choose can make a huge difference to the apparent image resolution. (Paint.net actually has an advantage here since it supports Fant, a very advanced and useful scaling algorithm that Photoshop and Gimp can't support.)

    • Like 1
    • Thanks 1
  17. 2 hours ago, Johnny Giotto said:

    You may be right about it having only one face.

    Oh, that at least is easy to find out. Drag a texture onto one spot of a copy of the mesh and see if it retextures the whole thing.

     

    2 hours ago, Johnny Giotto said:

    I just don't see how others are doing it.

    There's always the good old method of trial and error.

     

     

  18. 2 hours ago, arisaie said:

    I'm currently trying to make autumn leaves that are on top of my mesh.

    I'm not sure if I understand your question correctly but if I do, you're talking about a surface covered with leaves, not leaves on a tree, right?

    If so, here's something similar, a plastered wall covered with ivy:

    image.thumb.png.08ca3805ba04c21f9f9050ceb0047fb0.png

    The texture resolution isn't great but good enough. And it's a 512x512, if I really needed more, I'd use a 1024 instead:

    image.png.51d13c7a64d2f492f070dba89ba9c274.png

    The foliage is three dimensional with branches at different distances from the wall and often angled away from or towards the wall. That doesn't show up well in a picture so you have to take my word for it.

    It's a modluar system with different parts that can be combined in various ways to fit different wall sizes and shapes and give different amounts of coverage. This particular example has 356 vertices and 180 tris but it uses 13 modules (covering 12x5 m) so the LI couldn't be lower than 7 no matter how much I reduced the tri and vertice counts. If I were to make something similar as a single mesh, I could have reduced the mesh complexity significantly, possibly not all the way down to 1 for a mesh as big as this but certainly down to 2 or 3.

    The texture is fairly simple, it looks like this (except I scaled it down here, sorry I don't want to post full size copies of my textures on the forums):

    image.png.a2796db69557c393df65c6eb5110e4cc.png

    The trick is to use multiple overlapping panels instead of just a single big one and also to use texture repeats for all they're worth. Here's one of the modules:

    image.png.3e7613fb755196d051b4a0b9be46bf58.png

    This one only has two panels, so 8 vertices and 4 tris, but it has five instances of the texture with each panel covering the gaps in the other one:

    image.png.990d0c1bff681ca7144ba3a32dfb5fcc.png

    You get the idea? Getting it right with a good coverage and no visible repeats is a bit of an art both when it comes to texture creation and mesh making but it's not that difficult and it's sooo worth it!

    As for LI and LOD, you can make a 1 LI mesh with at least 6 single sided flat panels without reducing the LOD at all - use LOD above all the way. You should be able to get away with 10-12 and, depending on how they are configured, the alignment of the planets and the actions of a quantum butterfly in Inner Manchuria, you may be able to cram 20 or more panels into a single full LOD 1 LI mesh.

    • Like 1
  19. 4 hours ago, Kplh said:

    The texture is also removed from the faces of the primitive

    Yeah, that's an old one. I think what happened was that LL had to add this restriction to textures with restricted perms and either implemented it for full perm textures by accident or decided it was easier to implement it for all textures with no perm check. They never saw it as a big issue since there are at least two very obvious workarounds.

    • Like 1
    • Thanks 1
  20. 3 hours ago, Jennifer Boyle said:

    However, there were two problems. One was that the textures did not line up perfectly at the joints. The other was that fine lines were visible between the mesh faces.

    I would appreciate any comments. Could I have done something better? Are these issues inherent in SL?

    The fine lines between the mesh faces can have two different causes - or a combination of both.

    The first is a problem with texturing in general and not specifically SL related. There will always be a little bit of "bleeding" between opposite edges of a texture or UV island. (This is why it's usually recommended to keep a few pixels of space between the islands on a UV map btw.) One solution in SL is to reduce the texture repeat right a fraction, set them to 0.999 insted of 1.

    The second is that prim positions and sizes aren't always as precise as they should be in SL. No matter how well you align two prims, there may always be a tiny little gap between them. One possible solution is to increase the size of the prims so they overlap a little bit. Set the sizes to the lowest possible amount of overlap possible. There's no guarantee this will work but it's worth a try. A better solution would be to merge adjacent vertices in Blender but although this is a very simple process, I understand if you don't feel comfortable working in Blender at all.

    I have no real explanation for the messed up texture alignments. That's certainly not an SL issue so if the textures are aligned on the prim original it must be soemthing about Mesh Generator. One solution would be to use FIrestorm's mesh export function instead but then you would have to do some basic editing in Blender before you upload - no way around it at all.

    Of course, another alternative would be to keep the picture panel as prims. You only need three pathcut ones for it so the land impact will only be 2. If you decide to do it as six separate rpims, the LI will still only be 3. And of course, with prims the actual lag and load time will be (marginally) lower than if you convert it to mesh.

    Btw, if you decide to covert the panel to mesh, select "Use level above" for all LOD models. If you do the LOD models for a 12 tri mesh any other way than that, you're doing it wrong. There are a few equally good ways to do the physics model but the safe and easy one is to select "Cube" and click the the Analyze button.

    • Thanks 1
  21. 1 hour ago, IvyTechEngineer said:

    After all my effort it just didn't work.

    Two very quick examples. Ideally you want hull (that is analyzed) physcics for a wall but that's a little bit more complicated than surface (unanalyzed) physics for walls as complex as this and I'll try to keep it as simple as possible. For a build like yours it doesn't make that much of a difference anyway.

    For a archway/tunnel like this:

    image.png.70d28c0df9d064f8fa84f1f3c5ef0088.png

    You probably want a physics model like this, that is without the small polys along the edge:

    image.png.c06a572f17371c60d2009418539e3d4f.png

    Even this is probably too complex a physics model even though it won't cause any upload issues. Something like this would be even better:

    image.png.b82378ac1663cee75998ebfd520a156c.png

     

    ---

    Now for an arched hole in the wall:

    image.png.b7fec590fb920c0ad42c1f1e2bb62974.png

    Try this:

    image.png.4634bc12f71f6740c9e67d984d9ffeab.png

    or this:

    image.png.ce2b17e7a32138070f1f63ab584740aa.png

    or even this:

    image.png.6cbb9716f63502db4547baf1730eb1ee.png

    ---

    What is important to remember here, is that the only function the physics model for a wall has is to prevent avatar from walking through the wall. Avatar movements aren't very precise so you don't need detailed physics. And you probably don't need the physics model to cover the area above the opening since there's not much chance an avatar will ever end up up there.

    • Like 2
    • Thanks 2
  22. 14 hours ago, lookin4d said:

    It's going on 2 weeks or so at this point.

    That's a bit long and there may not be a backup copy of the sim anymore. But you should try to file a support ticket to LL about it. They won't give you your land back or refund your rent but in situations like this they are usually willing to reinstall the sim temporarily to give the tenants time to recover their items.

    • Like 4
  23. 51 minutes ago, Fluffy Sharkfin said:

    As for specific ideas on what those improvements and new features may be there have already been some great suggestions including a recent and quite detailed post by @ChinRey on ways in which the existing primitives system could be expanded and improved upon.

    Thank you for remembering. 🙂

    It wasn't that detailed a post though, only a few scattered examples and I didn't even mention what is possibly the most interesting way the prim system could be expanded: how prims could be merged into single objects without turning them into hard-to-handle polylist meshes.

    How to create a better in-world system is a bit off topic in this thread though, what is relevant here is why we should do it.

    The answer is obvious: It adds more activity to a virtual world.

    I know a guy who works as an IT teacher at a high school. He told me how impressed his students were when they saw the lovely home he had made in Meta's metaverse and how fast their enthusiasm evaporated when they realized the only thing they could actually do there was walk around and watch the scenery.

    All attempts at creating a shared virtual reality suffer from this problem, they tend to get boring fast because there isn't really that much to do there.

    In-world building for fun and profit (or at least the illusion of future profit) is one solution to this and it's fairly easy to implement, especially in Second Life. All game/VR platforms I'm aware of have some kind of in-world building options but none of them are anywhere near as advanced and flexible as SL's (and opensim's) prims and that gives us a very strong advantage over the competition. If only we could take advantage of it.

    • Like 3
    • Thanks 1
×
×
  • Create New...